Made gyp file for command buffer libraries....

Made gyp file for command buffer libraries.
Gyp build works on Windows. Mac and linux won't work yet.
Fixed some warnings.
Switched from dxerr.lib to dxerr9.lib.
Implemented a Texture::SetRect case for DXT compressed textures.
Fixed division by zero for zero stride vertex buffers.
TEST=none
BUG=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=26271

[rlp, 2009-09-15 00:52:48]

http://codereview.chromium.org/200127/diff/1/23
File command_buffer/client/cross/cmd_buffer_helper.h (right):

http://codereview.chromium.org/200127/diff/1/23#newcode149
Line 149: unsigned int get_;
Why can't this be left as CommandBufferOffset?

http://codereview.chromium.org/200127/diff/1/8
File command_buffer/service/cross/gapi_decoder.cc (right):

http://codereview.chromium.org/200127/diff/1/8#newcode475
Line 475: bool line_smooth = !!cmd::LineSmoothEnable::Get(enables);
Why the double negation? Is this faster than casting to a bool? Otherwise, the
cast might be a bit clearer.

http://codereview.chromium.org/200127/diff/1/3
File core/cross/command_buffer/texture_cb.cc (right):

http://codereview.chromium.org/200127/diff/1/3#newcode111
Line 111: size_t bytes_per_row = image::ComputePitch(format, src_width);
It looks like you could pull a bunch of the similar code out (bytes_per_row,
actual memcpy) and only change the number of rows calculation based on whether
the texture is compressed.

[apatrick, 2009-09-15 17:14:28]

http://codereview.chromium.org/200127/diff/1/23
File command_buffer/client/cross/cmd_buffer_helper.h (right):

http://codereview.chromium.org/200127/diff/1/23#newcode149
Line 149: unsigned int get_;
CommandBufferOffset is signed, which leads to errors when compared to unsigned
int. Maybe CommandBufferOffset should be unsigned?

http://codereview.chromium.org/200127/diff/1/8
File command_buffer/service/cross/gapi_decoder.cc (right):

http://codereview.chromium.org/200127/diff/1/8#newcode475
Line 475: bool line_smooth = !!cmd::LineSmoothEnable::Get(enables);
It's not faster. The problem is visual studio reports a "performance" warning
even if you use a cast. !!x is the least offensive way I know to get around it.
(0 || x) and (false || x) also work.

http://codereview.chromium.org/200127/diff/1/3
File core/cross/command_buffer/texture_cb.cc (right):

http://codereview.chromium.org/200127/diff/1/3#newcode111
Line 111: size_t bytes_per_row = image::ComputePitch(format, src_width);
On 2009/09/15 00:52:48, rlp wrote:
> It looks like you could pull a bunch of the similar code out (bytes_per_row,
> actual memcpy) and only change the number of rows calculation based on whether
> the texture is compressed.

Done.

[rlp, 2009-09-15 21:48:52]

LGTM

One small comment item below.

http://codereview.chromium.org/200127/diff/1/8
File command_buffer/service/cross/gapi_decoder.cc (right):

http://codereview.chromium.org/200127/diff/1/8#newcode475
Line 475: bool line_smooth = !!cmd::LineSmoothEnable::Get(enables);
On 2009/09/15 17:14:28, apatrick wrote:
> It's not faster. The problem is visual studio reports a "performance" warning
> even if you use a cast. !!x is the least offensive way I know to get around
it.
> (0 || x) and (false || x) also work.

Maybe put a small note (at least on the first one) as to that fact. Otherwise
it's a bit cryptic as to why you'd choose to do it this way.

[apatrick, 2009-09-15 21:52:12]

Committed.

http://codereview.chromium.org/200127/diff/1/8
File command_buffer/service/cross/gapi_decoder.cc (right):

http://codereview.chromium.org/200127/diff/1/8#newcode475
Line 475: bool line_smooth = !!cmd::LineSmoothEnable::Get(enables);
On 2009/09/15 21:48:52, rlp wrote:
> On 2009/09/15 17:14:28, apatrick wrote:
> > It's not faster. The problem is visual studio reports a "performance"
warning
> > even if you use a cast. !!x is the least offensive way I know to get around
> it.
> > (0 || x) and (false || x) also work.
> 
> Maybe put a small note (at least on the first one) as to that fact. Otherwise
> it's a bit cryptic as to why you'd choose to do it this way.

Done.